The association between different types of anthropomorphism of artificial intelligence service agents and user trust: A three-level meta-analysis

As artificial intelligence service agents (AISAs) become increasingly widespread, anthropomorphism is often proposed as a way to foster user trust, yet empirical findings remain mixed. Drawing on competing trust-building and trust-undermining mechanisms, we conducted a three-level meta-analysis of 211 effect sizes from 136 independent samples ( N = 46,071) to estimate the anthropomorphism–trust association and examine its boundary conditions. The average association was positive and robust to alternative dependence models ( r = 0.462, 95% CI [0.420, 0.502]), but varied substantially across settings (95% prediction interval [−0.131, 0.811]). It was considerably weaker for objectively specified design cues ( r = 0.222) than for perception-based measures ( r = 0.543), most of which captured adjacent trait attributions such as competence and warmth. Restricting outcomes to global trust, reliance, or willingness to depend yielded a similar estimate ( r = 0.475), although construct overlap on the outcome side cannot be ruled out. Cognitive and emotional anthropomorphism were more strongly associated with trust than physical anthropomorphism in aggregate, but this difference was no longer significant after adjusting for operationalization; among design cues, physical and emotional anthropomorphism did not differ significantly, and no design-based cognitive effect sizes were available. The association was stronger in Eastern than in Western samples and in survey than in experimental studies, whereas no statistically significant moderation was detected for gender composition or the prespecified service-context categories. These findings caution against reading the pooled association as a uniform effect of anthropomorphic design and suggest that anthropomorphic cues should complement, rather than substitute for, demonstrable competence and reliability in AISA design and deployment.

Authors

Institutions

Publication Details

Journal
Technological Forecasting and Social Change
Published
2026-11-18
DOI
https://doi.org/10.1016/j.techfore.2026.124959
Primary Topic
AI in Service Interactions
Type
article
Field-Weighted Citation Impact
0.00
Controls
|||
ALL TIME
JAN
FEB
MAR
APR
MAY
JUN
JUL
AUG
SEP
OCT
NOV
article

The association between different types of anthropomorphism of artificial intelligence service agents and user trust: A three-level meta-analysis

Wanghao Dong, Zhihui Cai, Xin Zhao, Sujie Meng et al.
2 links
Technological Forecasting and Social Change
AI in Service Interactions
article

The association between different types of anthropomorphism of artificial intelligence service agents and user trust: A three-level meta-analysis

Wanghao Dong, Zhihui Cai, Xin Zhao, Sujie Meng, Weijun Wang, Jie Zhang
article en

Abstract

As artificial intelligence service agents (AISAs) become increasingly widespread, anthropomorphism is often proposed as a way to foster user trust, yet empirical findings remain mixed. Drawing on competing trust-building and trust-undermining mechanisms, we conducted a three-level meta-analysis of 211 effect sizes from 136 independent samples ( N = 46,071) to estimate the anthropomorphism–trust association and examine its boundary conditions. The average association was positive and robust to alternative dependence models ( r = 0.462, 95% CI [0.420, 0.502]), but varied substantially across settings (95% prediction interval [−0.131, 0.811]). It was considerably weaker for objectively specified design cues ( r = 0.222) than for perception-based measures ( r = 0.543), most of which captured adjacent trait attributions such as competence and warmth. Restricting outcomes to global trust, reliance, or willingness to depend yielded a similar estimate ( r = 0.475), although construct overlap on the outcome side cannot be ruled out. Cognitive and emotional anthropomorphism were more strongly associated with trust than physical anthropomorphism in aggregate, but this difference was no longer significant after adjusting for operationalization; among design cues, physical and emotional anthropomorphism did not differ significantly, and no design-based cognitive effect sizes were available. The association was stronger in Eastern than in Western samples and in survey than in experimental studies, whereas no statistically significant moderation was detected for gender composition or the prespecified service-context categories. These findings caution against reading the pooled association as a uniform effect of anthropomorphic design and suggest that anthropomorphic cues should complement, rather than substitute for, demonstrable competence and reliability in AISA design and deployment.

Technological Forecasting and Social ChangeVol. 236
University of Manchester (GB), Central China Normal University (CN), Beijing City University (CN)
AI in Service Interactions
AI Navigator

Ask Laika to Summarize, Analyze, and Connect papers live on the map.

Summarize Papers & Methodologies

Extract key findings, datasets, and comparative methods across publications.

Benchmark Rankings & Visual Analytics

Rank top research institutions, authors, funders, topics, and journals by Field-Weighted Citation Impact (FWCI) and paper volume with instant charts.

Connect Distant Disciplines

Bridge topological clusters on the map to find hidden collaborative intersections.

Laika access coming soon